supervision by observation
play

Supervision by observation using inductive programming Jos - PowerPoint PPT Presentation

Supervision by observation using inductive programming Jos Hernndez-Orallo (project leader: Carlos Monserrat) Departament de Sistemes Informtics i Computaci, Universitat Politcnica de Valncia, Spain. jorallo@dsic.upv.es AAIP2015


  1. Supervision by observation using inductive programming José Hernández-Orallo (project leader: Carlos Monserrat) Departament de Sistemes Informàtics i Computació, Universitat Politècnica de València, Spain. jorallo@dsic.upv.es AAIP’2015 – Approaches and Applications of Inductive Programming, Dagstuhl Seminar 15442, October 25-30 1

  2. • Task automation vs. task supervision • Project SuPERVaSION • Application domain: surgical training Outline • Capture and representation: first ideas • Related works, expressions of interest 2

  3. Task automation vs. task supervision  One of the major applications of inductive programming is the automation of repetitive tasks from examples.  A significant progress has recently taken place.  Many problems still look too challenging for current techniques. 3

  4. Task automation vs. task supervision  Automatic task supervision when the superviser learns from expert examples and compares with an apprentice.  Don’t forget the other cup! 4

  5. Task automation vs. task supervision  Different problems (but related).  Key differences.  For task supervision:  In principle, there is no need of learning the task completely , just some key steps that can be traced and identified when a novice or an operator is doing it.  Several ways of solving the task are possible . The supervisor must be able to consider all types of solutions.  Correction and feedback is also a possibility in supervision. 5

  6. Project SUPERVaSION  SuPERVaSION : “Automated supervision by observation: pervasive technology for autonomous skill acquisition and procedure execution assistance” .  Funded as an Explora project (2015-2016) for risky, challenging ideas. “We envisage automated assistants that after the observation of how an expert performs a task are able to supervise whether other humans are performing the task correctly, also by observation”  Some IP applications, especially in the area of learning assistants and education, have addressed this kind of problem 6

  7. Project SUPERVaSION  Need and impact:  “Many tasks may be easily spoilt or may lead to suboptimal results by a human mistake that deviates from the procedure or the demonstration. A supervision system would be able to detect and advice the operator in real time”.  “Many tasks are learnt by humans more efficiently if these have continuous supervision and get meaningful comprehensible feedback”.  “Quality control and teaching planning could be improved significantly by a recollection of how the procedures are performed at a high level and the effect of the feedback over the operators”.  Coding supervisors manually is repetitive and expensive, and may not cover all the possible ways of carrying out the task. 7

  8. Application domain: surgical training  Training Minimally Invasive Surgery (Laparoscopes):  Students learn the procedure from a description and a demo.  Students must repeat the procedure several times.  Students learn faster with supervision.  Practice is done with a virtual simulator or a box-trainer.  Box-trainers are much cheaper and tactile feedback is real.  Virtual simulators are very expensive. Not enough availability for students.  Virtual simulators usually incorporate tasks and low-level supervision. 8

  9. Application domain: surgical training  Automatic supervision in this domain for virtual surgery simulators has been attempted in different ways:  Markov processes  String similarity (longest common subsequence algorithm):  “Automatic supervision of gestures to guide novice surgeons during training” C. Monserrat, A. Lucas, J. Hernández-Orallo, M. José Rupérez, Surgical Endoscopy (2014)  Using a character coding for gestures:  This is very low-level. Ignores the high-level description. 9

  10. Application domain: surgical training  Exercises have more information:  DESCRIPTION:  VIDEO: 10

  11. Application domain: surgical training  Approach with inductive programming  Provide start-up declarative knowledge about the domain,  Analyse these logs and suggest segmentation,  Identify groups and find repetitive structures,  Turn them into high-level actions that represent a program,  Possibly make them available to the expert,  Perform a similar approach for each trainee performing the same task, locating matches and mismatches, and  Producing high-level online feedback for the user. 11

  12. a 12

  13. Capture and representation  Scene capture  Using track dots in the trainer-box when recording.  Positions are analysed and recorded for the key objects.  High-level knowledge representation  Event-Action declarative languages, such as event calculus and variants. For instance,  Nikos Katzouris, Alexander Artikis, Georgios Paliouras “Incremental learning of event definitions with Inductive Logic Programming” Machine Learning 100:555 – 585, 2015.  Uses XHAIL: Ray, O. (2009). Nonmonotonic abductive inductive learning. Journal of Applied Logic , 7 (3), 329 – 340. 13

  14. Related works, expressions of interest  These (Explora) projects are meant to analyse a challenging problem and see whether the proposed approach is feasible.  If the initial analysis (and possible prototype) is successful they usually lead to larger consortia and projects. Similar projects, ideas, approaches and papers are welcome!  Or… it can be seen as a CHALLENGE for the IP community!  Anyhow, approach me during the coffee break! 14

Recommend


More recommend